Amyloid precursor proteins are constituents of the presynaptic active zone

Authors


Address correspondence and reprint requests to Walter Volknandt, Department of Molecular and Cellular Neurobiology, Biologicum, Goethe-University, Max-von-Laue-Str. 13, 60438 Frankfurt/Main, Germany. E-mail: volknandt@bio.uni-frankfurt.de

Abstract

The amyloid precursor protein (APP) and its mammalian homologs, APLP1, APLP2, have been allocated to an organellar pool residing in the Golgi apparatus and in endosomal compartments, and in its mature form to a cell surface-localized pool. In the brain, all APPs are restricted to neurons; however, their precise localization at the plasma membrane remained enigmatic. Employing a variety of subcellular fractionation steps, we isolated two synaptic vesicle (SV) pools from rat and mouse brain, a pool consisting of synaptic vesicles only and a pool comprising SV docked to the presynaptic plasma membrane. Immunopurification of these two pools using a monoclonal antibody directed against the 12 membrane span synaptic vesicle protein2 (SV2) demonstrated unambiguously that APP, APLP1 and APLP2 are constituents of the active zone of murine brain but essentially absent from free synaptic vesicles. The specificity of immunodetection was confirmed by analyzing the respective knock-out animals. The fractionation experiments further revealed that APP is accumulated in the fraction containing docked synaptic vesicles. These data present novel insights into the subsynaptic localization of APPs and are a prerequisite for unraveling the physiological role of all mature APP proteins in synaptic physiology.

image

We deciphered the precise subcellular localization of APP at the nerve terminal. We demonstrate that APP and its family members, APLP1 and APLP2, are constituents of the presynaptic active zone, albeit virtually absent in synaptic vesicles (SV). Our findings open new avenues for understanding the physiological role of the mature APP proteins at synaptic contacts, implying a function in the physiology of neurotransmitter release.

Abbreviations used
APLP1

amyloid precursor like protein 1

APLP2

amyloid precursor like protein 2

APP

amyloid precursor protein

APPs

amyloid precursor protein family members

BH

brain homogenate

DTT

dithiothreitol

LF

lower gradient fractions

PAGE

polyacrylamide gel electrophoresis

SDS

sodium dodecyl sulfate

SV2

synaptic vesicle protein 2

SV

synaptic vesicle

Syt-1

synaptotagmin-1

TBS

Tris-buffered saline

UF

upper gradient fractions

Amyloid precursor protein (APP) is a ubiquitously expressed type 1 transmembrane glycoprotein with a large N-terminal extracellular domain, anchored by a single membrane spanning α-helical region and a short cytoplasmic tail (Kang et al. 1987). In the brain, APP is expressed at high abundance predominantly in neurons and one of its proteolytic products is the main protein component of senile plaques typical of Alzheimer's disease (Glenner and Wong 1984; Masters et al. 1985; Guo et al. 2012). APP has several alternatively spliced isoforms, whereby the 695 amino acid isoform (APP695) is neuron specific (Sandbrink et al. 1994). The mammalian APP family consists of three members, APP and the APP-like proteins APLP1 and APLP2. Whereas APP and APLP2 are ubiquitously expressed in the vast majority of adult tissues, APLP1 is restricted to the nervous system (Slunt et al. 1994; Lorent et al. 1995). APP is cotranslationally inserted into the ER membrane and transported to the Golgi apparatus, where it becomes post-translationally modified by N- and O-glycosilation, sialylation, and by the attachment of glycosaminoglycans (Lyckman et al. 1998). APP is axonally sorted (Tienari et al. 1996) to a vesicular compartment. Using kinesin as a motor and microtubules as tracks, it reaches the presynapse by fast anterograde axonal transport where it becomes incorporated into the presynaptic plasma membrane (Lazarov et al. 2005; Kins et al. 2006; Szodorai et al. 2009). However, its precise localization remains elusive.

Employing subcellular fractionation of synaptosomes derived from rat brain and using a monoclonal antibody directed against the integral vesicle membrane protein synaptic vesicle protein 2 (SV2) we previously succeeded in immunopurifying free synaptic vesicles as well as a presynaptic compartment that contains the active zone with SVs docked to the presynaptic plasma membrane and lacks elements of the post-synaptic membrane. The immunoisolated fractions were analyzed by electron microscopy demonstrating that the methodology used is well suited to render unambiguous conclusions about constituents of the active zone which were identified by western blotting and mass spectrometry (Morciano et al. 2005; Burré et al. 2006).

In this study, we aimed to decipher the precise localization of APP, APLP1, and APLP2 at the presynapse. The presynaptic proteome controls neurotransmitter release as well as the short and long-term structural and functional dynamics of the nerve terminal. An exocytosis-competent subpopulation of SVs is docked via complex protein interactions to the presynaptic active zone (Rizzoli and Betz 2004, 2005). The docking process of SVs is mediated by the concerted action of three proteins, the vesicle localized synaptobrevin-2 (VAMP-2), the plasma membrane associated syntaxin-1A and SNAP25, forming a sodium dodecyl sulfate (SDS)-resistant quadruple helical coiled coil, the so-called SNARE complex (Montecucco et al. 2005). Here, we demonstrate that all three APP family members are constituents of the presynaptic active zone. Our findings open new avenues for understanding the physiological role of the mature APP proteins at synaptic contacts, implying a function in the physiology of neurotransmitter release.

Materials and methods

Animals

Animal treatment was performed under veterinary supervision according to European Guidelines. Wistar-rats of both sexes weighing 150 g to 225 g were obtained from Harlan-Winkelmann (Borchen, Germany) or from in-house breeding. C57BL/6N mice of both sexes and 8–10 weeks of age were obtained from Charles River (Sulzfeld, Germany) or from in-house breeding. Animals were kept under 12 h light and dark cycle with food and water ad libitum. The generation of mutant homozygous knockout mice APP–/–, APLP1–/–, and APLP2–/– has recently been described (Li et al. 1996; von Koch et al. 1997; Heber et al. 2000).

Antibodies

The two antibodies against APP were Y188 (rabbit monoclonal, 1 : 1000; Epitomics, Abcam, Burlingame, CA, USA) and 22C11 (mouse monoclonal, 1 : 1000, Millipore, Darmstadt, Germany). Additional antibodies used were directed against APLP1 (CT-11, rabbit polyclonal, 1 : 2000; Calbiochem, Darmstadt, Germany), APLP2 (D2-II, rabbit polyclonal, 1 : 2000, Calbiochem), Synaptotagmin-1 (rabbit polyclonal, 1 : 1000, SySy), SV2 (the clone CKK 10H4 producing the monoclonal anti-SV2 antibody, kindly donated by Dr. Regis B. Kelly, San Francisco, was cultured in-house). Dynabeads M-280 conjugated with monoclonal either sheep anti-mouse IgGs (cat. No. 112.02D) or sheep anti-rabbit IgGs (cat. No. 112.04D) were purchased from Invitrogen (Darmstadt, Germany).

Subcellular fractionation of free synaptic vesicles and the docked synaptic vesicle compartment from rat and mouse brain

Isolation of SVs and of the presynaptic membrane compartment containing the docked SVs from rat brain was performed essentially as previously described (Morciano et al. 2005). SVs were isolated from synaptosomes according to the protocol guidelines of Whittaker (Whittaker 1984). The protocol is similar to that used for rat brain but it had to be downscaled (DS-I). The following modifications were applied: Briefly, mouse brains were homogenized each in 1.5 mL of preparation buffer (5 mM Tris-HCl, 320 mM sucrose, pH 7.4) containing protease inhibitors as described for rat brain. The material was kept at 4°C during the preparation. The brain homogenate (BH) was centrifuged for 10 min at 1000 g. The resulting pellet (P1) was discarded and the supernatant (S1) was further fractionated by discontinuous Percoll gradient centrifugation. The Percoll gradient was prepared by layering 2.0 mL supernatant solution onto three layers of 2.0 mL each of Percoll solution [3, 10, 23% (v/v) in preparation buffer]. After centrifugation for 7 min at 32 500 g, fractions containing synaptosomes (Perc) were collected and diluted twofold in preparation buffer and centrifuged for 35 min at 44 000 g. For hypoosmotic shock of synaptosomes, the resulting pellet was resuspended in four volumes of lysis buffer (5 mM Tris-HCl, pH 7.4). The suspension was centrifuged for 1 h at 192 000 g. The pellet was resuspended and homogenized in 1.5 mL sucrose buffer (10 mM HEPES-NaOH, 0.5 mM EGTA, 0.1 mM MgCl2, 200 mM sucrose, pH 7.4). This microsomal solution (fraction P2) was layered onto 8.0 mL of a continuous sucrose gradient ranging from 0.3 M to 1.2 M sucrose (containing 10 mM HEPES-NaOH, 0.5 mM EGTA, pH 7.4) and centrifuged for 2 h at 85 000 g. Thirty-two fractions (300 μL each) were collected from top to bottom of the gradient. The pooled upper fractions (UF, 8 to 14, corresponding to 0.35–0.58 M sucrose) and the pooled lower fractions (LF) 22 to 30 corresponding to sucrose concentrations of 0.83 to 1.06 M were further analyzed.

Immunopurification of SVs and docked SV compartments from rat and mouse brain

Immunopurification of SVs in the pooled upper fractions (6–12) and of the active zone compartments in the pooled lower factions (24–32) was performed using a monoclonal antibody against the ubiquitous 12 span SV protein and putative transporter SV2 coupled to magnetic beads as previously described (Morciano et al. 2005). Elution of bound material was performed with 2% NP-40 (E1) in TBS for 30 min at 4°C, followed by a second elution step for 30 min at 22°C with sample buffer containing 2% SDS and 100 mM dithiothreitol (DTT) (E2).

Immunopurification of SVs from mouse brain was essentially performed as for rat brain-derived SVs with slight modifications. Briefly, 250 μL magnetic beads pre-coupled with an anti-mouse monoclonal antibody were washed with Tris-buffered saline (TBS, pH 7.4) and incubated with TBS containing 1% glycine, 1% lysine, and 0.5% saponin followed by three washing steps in TBS. Magnetic beads were then incubated for 1 h with the anti-SV2 antibody. Crosslinking of the antibodies was performed with 0.1% glutaraldehyde in TBS for 5 min and stopped by adding TBS containing 1% glycine and 1% lysine. Finally, beads were incubated over night at 4°C with the pooled upper (UF, 8–14) and lower sucrose gradient fractions (LF, 22–30). Beads containing the bound material were washed with TBS for three times and incubated with ice-cold acidified acetone (acetone containing 125 mM HCl) for 30 min at −20°C. Elution was performed with 2% NP-40 (E1) in TBS for 30 min supported by incubation (two times for 20 s) in an ultrasonic bath. Subsequently, a second elution step was performed with sample buffer containing 2% SDS and 100 mM DTT (E2).

Enrichment studies

For the analysis of the enrichment of specific protein components the BCA-assay kit (Pierce, Rockford, IL, USA) was used. Samples comprised BH, supernatant (S1), pellet (P1), crude synaptosomal fractions (Perc), crude microsomal fraction (P2), immunopurified docked SVs derived from the pooled LF eluted with 2% SDS (IP) and the remaining supernatant after immunoprecipitation (IP-S).

Quantification of immunosignals was performed with samples obtained under identical experimental conditions (= 4). Pixel intensities of bands were measured in voxels using ImageQuant TL software (GE Healthcare Life Sciences, Solingen, Germany) software from the same blot each. The starting material (BH) was set 100%.

Western blotting

Samples were run on a 10% Tris-glycine SDS-PAGE (Laemmli 1970) and transferred to nitrocellulose membrane (GE Healthcare Life Sciences, Solingen, Germany) using semi-dry blotting techniques (BioRad, Munich, Germany). Membranes were blocked with 5% milk in PBD/T phosphate-buffered saline containing Tween 20 (PBS/T) (123 mM NaCl, 7.4 mM Na2HPO4, 4.3 mM KH2PO4, 0.1% Tween20) for 1 h. Incubation with the respective primary antibody was performed over night at 4°C followed by a second blocking step with 5% skimmed milk powder (five times, 10 min each), subsequent incubation with the respective horseradish peroxidase-conjugated secondary antibody (GE Healthcare) and a final washing step in PBS/T (five times, 10 min each). Immunoblots were incubated with Western Lightning ECL substrate and visualized using ImageQuant LAS 4000 (both GE Healthcare).

Immunoprecipitation of APP

Immunoprecipitation was performed essentially as described for the monoclonal anti-SV2 antibody with the following modifications. The rabbit monoclonal antibody Y188 conjugated to pre-coupled sheep anti-rabbit IgG magnetic beads was incubated with the pooled lower sucrose density gradient fractions (LF) derived from mouse brain.

Electron microscopy

Fractions for electron microscopical analysis were prepared essentially as previously described (Morciano et al. 2005). Briefly, all solutions were prepared in phosphate-buffered saline (137 mM NaCl, 3 mM KCl, 13 mM Na2HPO4, 2 mM KH2PO4, pH 7.4) and all reactions were performed at 22°C unless stated otherwise. Gradient fractions were immunopurified as described above. After sample binding, beads were washed as described above, including treatment with Na2CO3, and embedded in 2% agarose (modified from De Camilli et al. 1983). The agarose cubes were fixed for 30 min in freshly prepared 3% formaldehyde and 0.25% glutaraldehyde or for 60 min in 2% glutaraldehyde. The embedded beads were washed in phosphate-buffered saline and post-fixed in 1% glutaraldehyde for 30 min and after renewed washing post-fixed for 2 h in 1% OsO4. After washing in phosphate-buffered saline, agarose blocks were dehydrated in a graded series of ethanol and propylene oxide and embedded in Epon (Fluka, Deisenhofen, Germany). Ultrathin sections were prepared using a Reichert Ultracut S ultra-microtome and stained with 5% uranyl acetate and lead citrate. For analysis, a Philips (Eindhoven, the Netherlands) CM 12 transmission electron microscope was used.

Results

The aim of our study was to decipher the precise subcellular localization of APP at the nerve terminal employing a variety of subcellular fractionation methods (Fig. 1). Following differential centrifugation steps synaptosomes derived from rat brain were enriched using Percoll gradient centrifugation. Subsequently, hypoosmotically shocked synaptosomes were subjected to continuous sucrose density gradient centrifugation. Upper and lower gradient fractions (UF and LF), respectively, were pooled and immediately subjected to immunopurification. Immunopurified material was subjected to western blotting and electron microscopy. Electron microscopic analysis of the immunopurified upper fractions yielded thousands of synaptic vesicle profiles (SV) immunolinked to the surface of the magnetic beads (MB). In contrast, the LF contained plasma membrane (PM)-attached synaptic vesicles bound to the magnetic beads (Fig. 1). Of note, we did not observe any patches of post-synaptic membranes attached to the presynaptic plasma membrane by electron microscopy.

Figure 1.

Flow-chart diagram of the experimental procedures. The immunopurified upper (UF) and lower fractions (LF) are highlighted by electron microscopic images. MB, magnetic beads; PM, plasma membrane; SV, synaptic vesicles.

Western blot analysis of every second fraction of the gradient revealed the migration behaviour of selected proteins in the density gradient. The sedimentation pattern of APP was compared with that of marker proteins of SVs. As revealed by immunoblotting, the bulk of the immunosignal for the APP migrated to the denser fractions of the sucrose gradient (Fig. 2a) with a peak at 0.83 M sucrose. Faint immunosignals were also detected in the lighter fractions in the upper part of the gradient. The band at 110 kDa represents the full-length neuronal APP as revealed by immunoblotting with a monoclonal antibody (22C11) directed against the amino acid residues sequence 66–81 within the N-terminal E1-domain and a monoclonal antibody (Y188) raised against the C-terminal YENPTY motif. The amyloid precursor-like proteins APLP1 and APLP2 revealed a similar migration pattern as APP (Fig. 2a). In contrast, the ubiquitous SV2 demonstrated a bimodal distribution with one peak in the upper fractions (UF, 0.29 M to 0.44 M sucrose) and a broad plateau in the LF (0.83 to 0.95 M sucrose). As previously shown by electron microscopy, immunodetection, and proteomics (Morciano et al. 2005, 2009) sucrose density gradient centrifugation enables us to distinguish between two SV pools. The separation of these two synaptic vesicle pools has recently been confirmed (Boyken et al. 2013). The upper fractions comprise free SVs and the LF contain docked SVs attached to the presynaptic active zone. The distribution of the vesicular calcium sensor synaptotagmin-1 (Syt-1) corresponds to that of SV2 and is present in both pools (Fig. 2a, bottom).

Figure 2.

Amyloid precursor protein (APP) family members are constituents of the active zone in murine brain. Migration in the sucrose density gradient of APP family members and synaptic vesicles (SV) marker proteins after loading of lysed rat brain synaptosomes. Every second fraction was immunoblotted (numbers on top). The bottom fraction was omitted. Bars on top (UF, upper fractions; LF lower fractions) indicate the peaks in the bimodal migration pattern of SV marker proteins representing free SVs and SVs docked to the active zone. (b) Western blot analysis of immunoprecipitated proteins from the pooled UF and LF comprising the free SV pool or SVs docked to the presynaptic active zone, respectively, eluted with NP-40 (E1) and subsequently with SDS and DTT (E2). Note that SV2 as the target for the monoclonal antibody used for immunoisolation could be detected in the second eluate only. 7.5 μL of eluate fractions were loaded per lane. (c) Migration in the sucrose density gradient of APP family members after loading of lysed mouse brain synaptosomes. Note the reduced distance between the two SV pools (cf. Fig. a). (d) Western blot analysis of immunoprecipitated material derived from mouse brain comprising the free SV pool (left) or SVs docked to the presynaptic plasma membrane (right) eluted by NP-40 (E1) and subsequently by SDS and DTT (E2). 7.5 μL of eluate fractions were loaded per lane. APP (22C11), monoclonal antibody directed against an amino acid sequence residing in the N-terminal region of the amyloid precursor protein; APP (Y188), monoclonal antibody against the motif in the C-terminus. APLP1 and APLP2, amyloid precursor like proteins 1 and 2, respectively. SV2, SV protein 2, and Syt-1, synaptotagmin-1. The bands correspond to 110 kDa for APP, 90 kDa for APLP1, 130 kDa for APLP2, 80 and 86 for SV2, and 65 for Syt-1.

To further purify free SVs and SVs docked to the presynaptic plasma membrane, the respective gradient fractions (indicated by the bars in Fig. 2a) were pooled and subjected to immunoisolation using a monoclonal antibody directed against the cytosolic N-terminus of the integral synaptic vesicle protein SV2 coupled to magnetic beads. To verify that the entire vesicle compartment becomes immunoprecipitated, we used synaptotagmin-1 as a synaptic vesicle marker protein. The pooled free SV fraction used as a starting material contained only traces of APP as revealed by immunodetection, whereas the majority of APP was present in the active zone fractions (cf. Fig. 2a). After immunoprecipitation, bound material was recovered by two consecutive elution steps employing first 100 μL of 2% non-ionic detergent NP-40 (Fig. 2b, E1) followed by a second elution with 100 μL sample buffer containing 2% SDS and 100 mM reducing agent (DTT) (Fig. 2b, E2). In western blotting, employing the pooled upper fractions (UF) as starting material, no immunosignal for APP could be detected in either eluate (Fig. 2b, UF); however after long exposure times, a faint signal could be detected in eluate fraction 2 (UF, E2). Moreover, a faint immunoreactive band for APLP1 could be observed in the sample buffer eluate (Fig. 2b, UF, E2). However, no immunosignal for APLP2 was obtained (Fig. 2b UF). In contrast, strong signals for all three APP family members could be immunodetected in the LF comprising SVs docked to the presynaptic active zone (Fig. 2b). As expected, the SV markers SV2 and synaptotagmin-1 (Syt-1) could be immunoprecipitated from both the pooled upper gradient fractions and the lower fractions (Fig. 2b). The second elution step in the presence of SDS and DTT yielded stronger immunosignals for all proteins analyzed. Of note, SV2, the target for the monoclonal antibody used for immunoisolation required SDS and DTT to be released into the eluate (Fig 2b, E2).

Previously, mouse mutants lacking individual or combinations of several APP family members were generated in mice (for review see Aydin et al. 2012). We therefore investigated the migration pattern of APP proteins in lysed synaptosomes derived from mouse brain. To account for the smaller mouse brains, we developed a downscaled experimental protocol, named DS-I. Applying this protocol and western blotting, we analyzed the migration behavior of the APP family members in fractions of lysed synaptosomes subjected to continuous sucrose density gradient centrifugation. APP migrated to the denser fractions of the sucrose gradient as it had been observed for rat brain (Fig. 2c, top; cf. Fig. 2a). Similarly, APLP1 and APLP2 migrated to the lower fractions (Fig. 2c). As for the rat, the ubiquitous synaptic vesicle proteins SV2 and synaptotagmin-1 revealed a bimodal distribution with one peak in the upper fractions (0.35 M to 0.58 M sucrose) and a broad plateau in the lower fractions (0.83 to 1.06 M sucrose), compatible with the migration behavior of free SVs and SVs docked to the presynaptic active zone. Downscaling to smaller sucrose gradient volumes reduced the resolution between free SVs and active zone-attached SVs by four fractions (Fig. 2c).

Immunoprecipitation of mouse brain synaptic vesicles from the pooled upper and lower fractions yielded results similar to those obtained for rat brain (Fig 2d, cf Fig. 2b). A faint immunosignal for APP could be visualized only after prolonged exposure times, indicating that the amount of APP is close to the detection limit. Again, no immunosignal for APLP2 and only a weak signal for APLP1 could be detected in eluate 2 (E2) when free SVs were used as starting material (Fig. 2d, UF). On the contrary, strong signals for all APP family members were obtained for the active zone fraction. Again, SV2 eluted preferentially in the presence of SDS and DTT.

The results obtained so far by sucrose density gradient centrifugation and immunopurification demonstrated that APPs are enriched in fractions containing SVs docked to the presynaptic plasma membrane. We further investigated whether APP becomes enriched in this fraction during consecutive purification steps. We compared the proteinaceous content from BH as the starting material, the first supernatant fraction used for further purification (S1), the respective pellet fraction (P1), the crude synaptosomal fraction (Perc), the microsomal fraction (P2), the immunopurified docked SVs attached to the presynaptic zone (IP), and finally the remaining supernatant after IP (IP-S) (Fig. 3a). Western blot analyses revealed a continuous increase in APP immunoreactivity during the course of subcellular fractionation. Importantly, immunoprecipitation of SVs attached to the active zone resulted in a five to sixfold enrichment of APP as compared to brain homogenate calculated in voxels (volumetric pixels) using ImageQuant TL (= 4). Accumulation of APP is accompanied by a simultaneous enrichment of synaptotagmin-1 (Fig. 3a). Preliminary data indicate a similar accumulation of APLP1 and APLP2 at the active zone (not shown).

Figure 3.

Amyloid precursor protein (APP) is highly enriched at the active zone. (a) Western blot analysis of material derived from mouse brain obtained after consecutive purification steps during the isolation procedure. BH, brain homogenate; S1, supernatant 1, P1 pellet 1; Perc, synaptosomal enriched fraction after Percoll gradient centrifugation; P2, pellet 2; IP, immunopurified eluate using SDS; IP-S, remaining supernatant after IP. 150 ng of protein was loaded per lane. All fractions were analyzed in the same blot for APP and synaptotagmin-1. The thin line indicates the excision of fractions from the compound blot. (b) Western blot analysis of immunoprecipitated material derived from synaptic vesicles (SVs) docked to the plasma membrane. Immunoprecipitate derived from mutant homozygous knockout (ko) mice APP−/−, APLP1−/−, and APLP2−/− and wildtype (wt) was loaded. Note, there is a faint immunosignal because of a cross-reactivity of the secondary anti-mouse antibody used in immunodetection when the mouse IgG was used for immunopurification. This is not seen when rabbit IgGs are used for immunodetection. (c) APP can be immunoprecipitated from the lower sucrose density gradient fractions derived from wild-type mouse (wt) brain using the Y188 antibody. No immunosignal was detected in APP−/− knockout mice. For immunodetection, the mouse monoclonal antibody directed against APP 22C11 was used. Antibodies and molecular masses of proteins as for Fig. 2.

To evaluate the specificity of antibodies used in our study we performed a comparative Western blot analyses of the immunopurified active zone fraction of wild type and homozygous knockout mice APP−/−, APLP1−/−, and APLP2−/− (Li et al. 1996; von Koch et al. 1997; Heber et al. 2000). Homozygous knockout was verified by genotyping. Immunoblotting using the monoclonal mouse APP antibody (22C11) directed against the amino acid 66–81 in the E1 domain, the monoclonal rabbit APP antibody (Y188), and the polyclonal rabbit antibodies detecting APLP1, and APLP2, respectively, revealed positive immunosignals in the wild-type fraction and the absence of the protein bands in the respective knockout mice (Fig. 3b), proving the specificity of the antibodies. The faint immunosignal in Fig. 3b (ko, 22C11) is because of a cross-reactivity of the secondary anti-mouse antibody that was used for immunodetection with the mouse IgG that was used for immunopurification. Importantly, however, this band is not observed when rabbit IgGs are used for immunodetection (Fig. 3b, ko, Y188).

To further corroborate the localization of APP at the active zone, we directly immunoprecipitated APP from the pooled lower sucrose density gradient fractions without previous immunoisolation, using the rabbit monoclonal antibody Y188. Subsequent immunodetection using the monoclonal anti-mouse APP antibody 22C11 confirmed that APP is a constituent of this fraction whereas it is absent in knockout animals (Fig. 3c).

Discussion

Using immunoisolation and APP-specific antibodies, we show that APP and its homologs APLP1 and APLP2 are constituents of the active zone proteome but not or only to a minor extent of free synaptic vesicles.

In the CNS, the APP is primarily expressed in neurons (Guo et al. 2012). Previous studies had attributed APP to the cell surface despite its presence in intracellular pools such as the Golgi apparatus and endosomes. An early report of Shivers and co-workers (Shivers et al. 1988) describes the appearance of the protein in patches on or near the plasma membranes of neurons. Post-translationally modified full-length APP is a component of the mature presynapse (Lyckman et al. 1998). It has been reported that full length APP is endocytosed from the cell surface (Lyckman et al. 1998; Cirrito et al. 2008) and sorted into endosomes (Ikin et al. 1996). In this context, it is noteworthy that marker proteins of endocytosis such as clathrin and dynamin also migrate to the denser fractions of the gradient thereby resembling the migration behavior of APP family members (Morciano et al. 2009). Moreover, subcellular fractionation of microsomes derived from NGF-differentiated PC12 cells revealed the enrichment of the early endosomal marker rab5 in the denser fractions (Barth et al. 2011). A recent report indicated that a small number of SVs contain APP (Groemer et al. 2011). This would be compatible with our observation that a faint immunosignal for APP could be visualized after prolonged exposure times.

Importantly, our observations reveal that not only APP but in addition its mammalian homologs APLP1 and APLP2 are constituents of the presynaptic active zone. In this context, it is of relevance that after immunopurification of SVs docked to the active zone, the post-synaptic density proteins PSD93, PSD95, and synapse-associated protein SAP102 could not be detected in this fraction; no immunosignals were obtained for the ER, Golgi complex, the early endosomal compartment, of lysosomes or of peroxysomes (Morciano et al. 2009). No post-synaptic components were identified by mass spectrometry supporting the evidence obtained by immunodetection (Morciano et al. 2009).

The close proximity of the APP family members to docked SVs could favor physical interactions. At the presynaptic active zone APPs might act in concert with a variety of other proteins in, for example, signal transduction and fine-tuning the cellular response. At the cell surface cis– or trans-homo- and heterointeractions (Soba et al. 2005) of all APPs have been postulated (Kins et al. 2006). In this context, it is noteworthy that the active zone contains numerous cell adhesion molecules (Morciano et al. 2009).

Analysis of APP-deficient mice revealed deficits in hippocampal long-term potentiation (LTP) and in short-term plasticity at neuromuscular synapses because of aberrant activation of presynaptic N-type and L-type calcium channels (Ring et al. 2007; Yang et al. 2007). Moreover, neuromuscular junctions of APP/APLP2 double knockout (DKO) mice showed a reduced density of SVs and impaired neurotransmission (Wang et al. 2005, 2009). Numerous presynaptic interactors of APP involved in vesicle transport, trafficking, and endocytosis have recently been identified (Reinhard et al. 2005; Anliker and Müller 2006; Wolfe and Guenette 2007). In embryonic stem cell-derived DKO, neurons expression of the vesicular glutamate transporter 2 was decreased (Schrenk-Siemens et al. 2008). Analysis of Munc 13-1 knock-out and knock-in mice revealed that Munc-13-1-mediated vesicle priming contributes to the regulation of APP metabolism (Rossner et al. 2004). Other constituents of the SV proteome (Burré et al. 2006) including dynamin-1, Munc18, N-ethylmaleimide sensitive fusion protein (Cottrell et al. 2005), SV2A (Norstrom et al. 2010), and synaptotagmin-1 (Kohli et al. 2012), might interact directly with APP (Volknandt and Karas 2012). Additional interactions partners might include synapsin 2, the vesicular proton-pump vATPase (Norstrom et al. 2010), and members of the 14-3-3 family (Kohli et al. 2012). These putative targets of APP are depicted in Fig. 4. It is noteworthy that APP is capable to form a complex with the active zone proteins Mint1, CASK (Wang et al. 2009), and bassoon (Norstrom et al. 2010).

Figure 4.

Cartoon emphasizing the presence of all three amyloid precursor protein (APP) family members (APP, APLP1, and APLP2) at the active zone and highlighting some of their putative targets as described by Cottrell et al. (2005), Kohli et al. (2012), Norstrom et al. (2010), Wang et al. (2009). SV, docked synaptic vesicle. Drawings are not to scale.

Our data unambiguously allocate all three APPs to the presynaptic active zone. This represents a prerequisite for further unravelling the interaction of APP family members with their targets and gaining novel insights into the physiological role of all mature APP proteins in synaptic physiology.

Acknowledgements

Financial support was provided by the Deutsche Forschungsgemeinschaft (VO423/12-1 to Walter Volknandt and MU 1457/8-1; 1457/8-2 to Ulrike Müller) and by grants from the Cluster of Excellence EXC 115 and Gutenberg Research College (GRC) University Mainz (to Amparo Acker-Palmer). We thank Marco Morciano for providing electron microscopical images. We are grateful to Herbert Zimmermann for valuable suggestions. We have no conflict of interest.

Ancillary